Research Groups

• Speech Information Processing Group (Associate Professor Takeshi Yamada)

  Our group studies on speech information processing focused on real environments, including noise robust speech recognition, usability improvement of noisy speech recognition, and objective quality evaluation of noise reduction algorithms. We recently give high priority to the development of an automatic scoring system for the Japanese speaking test SJ-CAT.
  
  For more details and publications, please visit Assoc. Prof. Yamada's website. [link]
  
  
  
  
  
  

• Acoustic Signal Processing Group (Visiting Professor Shoji Makino)

  The study of acoustic signal processing group covers wide area of audio signal processing to extend and improve various communication employing sound.
  We are usually surrounded by various sound except in special environments such as sound proof room. Also, there are various factors that degrade quality of sound recording such as reflection and reverberation in acoustic propagation systems and disturbances in electrical circuits. Thus, high-quality acquisition of specific sound sources requires special skills of recording engineers and special equipments. Such difficulty of sound recording itself is a major barrier of many technologies using sound, such as hands-free speech recognition/telecommunication, robot audition, music information processing, etc. The goal of the group is to develop technologies for decoding of information from degraded audio signal, signal restoration, and separation/emphasis of specific sound sources, with the help of the ideas brought by pattern recognition. The current main interest of the group includes:
  • Emphasis/separation of desired source from mixed observation of multiple sources
  • Source localization using microphone array
  • Asynchronous/distributed microphone array combining multiple mobile recording devices
  • Traffic monitoring based on acoustic sensing
  
  For more details and publications, please visit Prof. Makino's website. [link]